Skip to content

Latest commit

 

History

History
69 lines (57 loc) · 2.34 KB

File metadata and controls

69 lines (57 loc) · 2.34 KB

TensorFlow Build Dependencies Summary

Updated Dependencies (June 2025)

Core Versions

  • TensorFlow: 2.19.0 (latest stable)
  • Bazelisk: 1.26.0 (latest)
  • NumPy: 2.2.6 (latest compatible)
  • Six: 1.17.0 (latest)

System Dependencies

Essential Build Tools

  • curl - Download Bazelisk and other resources
  • git - Clone TensorFlow repository
  • pip - Python package management
  • build-essential - GCC, make, and other compilation tools
  • patchelf - ELF binary modification (required by TensorFlow)
  • python3-full - Complete Python installation
  • python3-dev - Python development headers
  • python3-distutils - Python distribution utilities
  • python3-venv - Virtual environment support

Mathematical Libraries (NEW)

  • libomp-dev - OpenMP for parallel processing
  • libopenblas-dev - Optimized BLAS implementation
  • liblapack-dev - Linear Algebra PACKage
  • libeigen3-dev - C++ template library for linear algebra
  • libblas-dev - Basic Linear Algebra Subprograms
  • libatlas-base-dev - Automatically Tuned Linear Algebra Software

System Tools

  • cpuinfo - CPU information utility (for core count detection)
  • llvm-17 - LLVM compiler infrastructure
  • clang-17 - Clang C/C++ compiler

New Features

OS Version Checking

  • Automatic detection of Ubuntu/Debian versions
  • Compatibility validation (Ubuntu 20.04+ recommended)
  • User confirmation for unsupported systems

Enhanced Mathematical Performance

The added mathematical libraries provide:

  • OpenMP: Parallel processing capabilities
  • OpenBLAS: Highly optimized BLAS routines
  • LAPACK: Advanced linear algebra operations
  • Eigen3: High-performance C++ linear algebra
  • ATLAS: Automatically tuned mathematical libraries

Dependency Analysis

All dependencies are actively used:

  • No redundant packages identified
  • Mathematical libraries complement each other
  • Each provides specific optimizations for TensorFlow

Testing

  • All dependencies verified through automated tests
  • OS compatibility checking implemented
  • Package installation validation included
  • CPU feature detection confirmed

Compatibility

  • OS: Ubuntu 20.04+ (primary), Debian 12+ (supported)
  • Python: 3.9-3.12 (tested with 3.12)
  • CPU: AVX2 and FMA support required
  • Memory: 8GB minimum, 16GB recommended
  • Storage: 50GB free space minimum